Public display screen content detection method and system based on video watermark verification

Abstract

The invention relates to a public display screen content detection method and system based on video watermark verification, and belongs to the technical field of information security. The method comprises the following steps: acquiring image data and audio data of a to-be-played image, and performing preprocessing according to the image data and the audio data to obtain image preprocessing data and sound preprocessing data; performing identification according to the image preprocessing data and the sound preprocessing data to obtain an abnormal identification result and a sound abnormal identification result, and performing calculation through an abnormal processing model according to the abnormal identification result and the sound abnormal identification result to obtain safety image data; watermark information is preset, the encryption server embeds the watermark information into the secure image data through an SM2 algorithm to obtain encrypted image data and transmits the encrypted image data to the decryption server, the decryption server obtains an identification result through a decryption model according to the encrypted image data, and the player plays a video according to the identification result. And public display screen content security detection is realized.

Description

Technical Field

[0001] The present invention belongs to the technical field of information security, and in particular relates to a method and system for detecting content on a public display screen based on video watermark verification. Background Art

[0002] With the rapid development of urban digital transformation, various public display screens have greatly improved the efficiency of urban appearance construction and publicity and promotion business.

[0003] However, the current technology of public display screens is moving towards splicing technology and display technology, but ignores the issue of whether the broadcast content is compliant by the users in the later stage, resulting in a lack of safety awareness in the early stage of the industry. Therefore, the screen itself does not have the ability to review content security, which also leads to a series of content security broadcast control issues. Summary of the invention

[0004] In order to solve the above problems existing in the prior art, the present invention provides a method and system for detecting content on a public display screen based on video watermark verification.

[0005] The purpose of the present invention can be achieved by the following technical solutions:

[0006] S1: Acquire image data and audio data of a to-be-played image, obtain image preprocessing data by image preprocessing according to the image data, and obtain sound preprocessing data by sound preprocessing according to the audio data;

[0007] S2: Obtain an abnormality recognition result through an abnormality recognition model according to the image preprocessing data, obtain a sound abnormality recognition result through a sound abnormality recognition model according to the sound preprocessing data, and calculate the safety image data through an abnormality processing model according to the abnormality recognition result and the sound abnormality recognition result;

[0008] S3: preset watermark information, the encryption server embeds the watermark information into the secure image data through the SM2 algorithm to obtain encrypted image data, and transmits the encrypted image data to the decryption server, the decryption server obtains the recognition result according to the encrypted image data through the decryption model, and the player plays the video according to the recognition result.

[0009] Specifically, the anomaly recognition model includes an image anomaly recognition model and a text anomaly recognition model. The image anomaly recognition model includes:

[0010] S201: Frame-dividing the image preprocessing data to obtain an image sequence, sequentially traversing the image sequence to obtain a test frame, calculating a feature map using an IMTCNN model based on the test frame, and calculating a feature compensation factor using an operator based on the feature map;

[0011] S202: Calculate an optimal feature compensation coefficient using a GA algorithm according to the feature compensation factor, and calculate a feature description image according to the optimal feature compensation coefficient and the feature compensation factor;

[0012] S203: Preset a segmentation template, and segment the feature description image according to the segmentation template to obtain segmentation sub-images;

[0013] S204: Calculate the pixel value of the segmented sub-image, obtain the coordinates of the segmented sub-image, and splice each pixel value according to the coordinates to obtain a new histogram;

[0014] S205: According to the new histogram, dimension reduction is performed by using the PCA algorithm to obtain main characteristic factors, and according to the main characteristic factors, an abnormal image recognition result is calculated by using the SVM algorithm.

[0015] Specifically, the text anomaly recognition model includes:

[0016] S301: locating text using an SSD algorithm according to the test frame to obtain a text recognition area;

[0017] S302: performing text recognition according to the text recognition area by using a CRNN algorithm to obtain a text recognition result;

[0018] S303: performing text preprocessing according to the text recognition result to obtain a preprocessed text, wherein the text preprocessing includes data cleaning, noise removal, and stop word removal;

[0019] S304: A first-level traversal result is obtained by calculating through a first-level filter according to the preprocessed text, and the first-level traversal result is judged. If the first-level traversal result contains sensitive words, the preprocessed text is passed to the second-level filter, and the second-level filter calculates according to the preprocessed text to obtain a text abnormality recognition result; if the first-level traversal result does not contain sensitive words, the calculation is stopped and the text abnormality recognition result is output.

[0020] Specifically, the specific calculation steps of the abnormal sound recognition model are:

[0021] S401: Obtain an energy spectrum by fast Fourier transform according to the sound preprocessing data, and obtain an amplitude spectrum by removing the phase spectrum according to the energy spectrum;

[0022] S402: Obtain FBank features through Mel filtering according to the amplitude spectrum, obtain logarithmic features through logarithmic calculation according to the FBank features, and obtain MFCC features through discrete cosine transform according to the logarithmic features;

[0023] S403: Obtaining speech text data by calculating the MFCC features through a DNN-LSTM model;

[0024] S404: A first-level traversal result is obtained by calculating through a first-level filter according to the voice text data, and the first-level traversal result is judged. If the first-level traversal result contains sensitive words, the preprocessed text is passed to the second-level filter, and the second-level filter calculates according to the voice text data to obtain the sound abnormality recognition result; if the first-level traversal result does not contain sensitive words, the calculation is stopped and the sound abnormality recognition result is output.

[0025] Specifically, the abnormality processing model includes making a judgment according to the abnormality recognition result and the abnormal sound recognition result, and if the abnormality recognition result and the abnormal sound recognition result are normal, outputting the image to be played;

[0026] If the abnormal identification result is abnormal, the test frame is transmitted to the human-computer interaction page, and a confirmation signal is obtained through the human-computer interaction interface, and the confirmation signal is judged. If the confirmation signal is confirmed to be abnormal, a replacement frame is preset, and the replacement frame is obtained through the human-computer interaction interface, and the replacement frame is replaced with the test frame to obtain the safe image; if the confirmation signal is confirmed to be normal, the image to be played is output;

[0027] If the sound abnormality recognition result is abnormal, the sound pre-processing data is passed to the human-computer interaction page, and the confirmation signal is obtained through the human-computer interaction interface, and the confirmation signal is judged. If the confirmation signal is a confirmation of abnormality, a replacement voice is preset, and the replacement voice is obtained through the human-computer interaction interface. The replacement voice is replaced with the sound pre-processing data to obtain the safety image; if the confirmation signal is a confirmation of normal, the image to be played is output.

[0028] Specifically, the decryption server obtains a recognition result according to the encrypted image data through a decryption model, and the player plays the video according to the recognition result, which specifically includes:

[0029] Decrypting the encrypted image data using the SM2 algorithm to obtain hidden watermark information, and comparing the hidden watermark information with the watermark information to obtain the recognition result, wherein the recognition result includes the same and different;

[0030] The player judges the recognition result, and if the recognition result is the same, the encrypted image data is played; if the recognition result is different, the image playback is stopped.

[0031] Specifically, the image preprocessing includes filtering noise reduction, histogram equalization, grayscale, data enhancement, light compensation, and geometric correction; the sound preprocessing includes noise reduction, dereverberation, pre-emphasis, and frame windowing.

[0032] A public display screen content detection system based on video watermark verification, comprising: a preprocessing module, an identification module, and a decryption module;

[0033] The preprocessing module is used to obtain image data and audio data of the image to be played, obtain image preprocessing data by performing image preprocessing according to the image data, and obtain sound preprocessing data by performing sound preprocessing according to the audio data;

[0034] The recognition module is used to obtain an abnormality recognition result through an abnormality recognition model according to the image preprocessing data, obtain a sound abnormality recognition result through a sound abnormality recognition model according to the sound preprocessing data, and calculate the safety image data through an abnormality processing model according to the abnormality recognition result and the sound abnormality recognition result;

[0035] The decryption module is used to preset watermark information. The encryption server embeds the watermark information into the secure image data through the SM2 algorithm to obtain encrypted image data, and transmits the encrypted image data to the decryption server. The decryption server obtains a recognition result based on the encrypted image data through a decryption model, and the player plays the video based on the recognition result.

[0036] An electronic device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method for detecting content of a public display screen based on video watermark verification as described above is implemented.

[0037] A storage medium containing computer executable instructions, which are used to perform the public display screen content detection method based on video watermark verification as described above when executed by a computer processor.

[0038] The beneficial effects of the present invention are:

[0039] (1) By setting up an anomaly recognition model and an anomaly recognition model for sound, the video is identified for anomalies in terms of pictures, text, and language. This avoids the appearance of prohibited pictures, words, and voices in the video while using neural networks for identification, thus solving problems such as detection errors caused by manual detection methods.

[0040] (2) By setting up the SM2 encryption algorithm and presetting the encryption information, the encrypted information is added to the video in the form of a hidden watermark through the SM2 algorithm. The decoding server decrypts the information through the SM2 algorithm and verifies whether the watermark information is the same. The video is played after the verification is passed, thereby avoiding broadcast accidents caused by others arbitrarily replacing the video. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to facilitate understanding by those skilled in the art, the present invention is further described below with reference to the accompanying drawings.

[0042] Figure 1 The present invention is a flowchart of a method for detecting content on a public display screen based on video watermark verification. DETAILED DESCRIPTION

[0043] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation methods, structures, features and effects of the present invention are described in detail below in conjunction with the accompanying drawings and preferred embodiments.

[0044] See also Figure 1 , a method and system for detecting content on a public display screen based on video watermark verification,

[0045] S1: Acquire image data and audio data of a to-be-played image, obtain image preprocessing data by image preprocessing according to the image data, and obtain sound preprocessing data by sound preprocessing according to the audio data;

[0046] S2: Obtain an abnormality recognition result through an abnormality recognition model according to the image preprocessing data, obtain a sound abnormality recognition result through a sound abnormality recognition model according to the sound preprocessing data, and calculate the safety image data through an abnormality processing model according to the abnormality recognition result and the sound abnormality recognition result;

[0047] S3: preset watermark information, the encryption server embeds the watermark information into the secure image data through the SM2 algorithm to obtain encrypted image data, and transmits the encrypted image data to the decryption server, the decryption server obtains the recognition result according to the encrypted image data through the decryption model, and the player plays the video according to the recognition result.

[0048] Specifically, the image preprocessing includes filtering noise reduction, histogram equalization, grayscale, data enhancement, light compensation, and geometric correction.

[0049] Specifically, the sound preprocessing includes noise reduction, dereverberation, pre-emphasis, and frame windowing.

[0050] Specifically, the anomaly recognition model includes an image anomaly recognition model and a text anomaly recognition model. The image anomaly recognition model includes:

[0051] S201: Frame-dividing the image preprocessing data to obtain an image sequence, sequentially traversing the image sequence to obtain a test frame, calculating a feature map using an IMTCNN model based on the test frame, and calculating a feature compensation factor using an operator based on the feature map;

[0052] S202: Calculate an optimal feature compensation coefficient using a GA algorithm according to the feature compensation factor, and calculate a feature description image according to the optimal feature compensation coefficient and the feature compensation factor;

[0053] The feature description image calculation formula is:

[0054]

[0055] Among them, sig is the feature description image, A is the test frame, and F i is the optimal characteristic compensation coefficient, S i is the characteristic compensation factor, and n is the number of characteristic compensation factors.

[0056] S203: Preset a segmentation template, and segment the feature description image according to the segmentation template to obtain segmentation sub-images;

[0057] S204: Calculate the pixel value of the segmented sub-image, obtain the coordinates of the segmented sub-image, and splice each pixel value according to the coordinates to obtain a new histogram;

[0058] S205: According to the new histogram, dimension reduction is performed by using the PCA algorithm to obtain main characteristic factors, and according to the main characteristic factors, an abnormal image recognition result is calculated by using the SVM algorithm.

[0059] It should be noted that the operators include left bias matrix, right bias matrix, upper bias matrix, lower bias matrix, difference matrix, fusion matrix, feature map denoising, and function amplification.

[0060] Specifically, the text anomaly recognition model includes:

[0061] S301: locating text using an SSD algorithm according to the test frame to obtain a text recognition area;

[0062] S302: performing text recognition according to the text recognition area by using a CRNN algorithm to obtain a text recognition result;

[0063] S303: performing text preprocessing according to the text recognition result to obtain a preprocessed text, wherein the text preprocessing includes data cleaning, noise removal, and stop word removal;

[0064] S304: A first-level traversal result is obtained by calculating through a first-level filter according to the preprocessed text, and the first-level traversal result is judged. If the first-level traversal result contains sensitive words, the preprocessed text is passed to the second-level filter, and the second-level filter calculates according to the preprocessed text to obtain a text abnormality recognition result; if the first-level traversal result does not contain sensitive words, the calculation is stopped and the text abnormality recognition result is output.

[0065] It should be noted that the abnormality recognition result includes the image abnormality recognition result and the text abnormality recognition result. The image abnormality recognition result includes a normal picture and an abnormal image, and the text abnormality recognition result includes a normal text and an abnormal text.

[0066] In this embodiment, the first-level filter calculation process includes: scanning a preset sensitive word library, constructing a dictionary tree according to the preset sensitive word library, and sequentially traversing the preprocessed text according to the dictionary tree to obtain a first-level traversal result; the second-level filter calculation process includes: obtaining a segmented text through Jieba segmentation according to the preprocessed text, calculating a text feature vector through a bag of words model according to the segmented text, and calculating the text anomaly recognition result through a classifier according to the text feature vector.

[0067] Specifically, the specific calculation steps of the abnormal sound recognition model are:

[0068] S401: Obtain an energy spectrum by fast Fourier transform according to the sound preprocessing data, and obtain an amplitude spectrum by removing the phase spectrum according to the energy spectrum;

[0069] S402: Obtain FBank features through Mel filtering according to the amplitude spectrum, obtain logarithmic features through logarithmic calculation according to the FBank features, and obtain MFCC features through discrete cosine transform according to the logarithmic features;

[0070] S403: Obtaining speech text data by calculating the MFCC features through a DNN-LSTM model;

[0071] S404: A first-level traversal result is obtained by calculating through a first-level filter according to the voice text data, and the first-level traversal result is judged. If the first-level traversal result contains sensitive words, the preprocessed text is passed to the second-level filter, and the second-level filter calculates according to the voice text data to obtain the sound abnormality recognition result; if the first-level traversal result does not contain sensitive words, the calculation is stopped and the sound abnormality recognition result is output.

[0072] It should be noted that the abnormal sound recognition result includes normal sound and abnormal sound.

[0073] In this embodiment, the phase spectrum of the energy spectrum is removed by calculating the absolute value of each complex number in the energy spectrum, and removing the energy spectrum according to the absolute value to obtain the amplitude spectrum;

[0074] The DNN-LSTM model uses an LSTM neural network as the bottom layer and a DNN neural network as the high layer. The LSTM neural network uses a sigmoid activation function, while the DNN neural network uses a sigmoid activation function and a tanh function in combination.

[0075] Specifically, the abnormality processing model includes making a judgment according to the abnormality recognition result and the abnormal sound recognition result, and if the abnormality recognition result and the abnormal sound recognition result are normal, outputting the image to be played;

[0076] If the abnormal identification result is abnormal, the test frame is transmitted to the human-computer interaction page, and a confirmation signal is obtained through the human-computer interaction interface, and the confirmation signal is judged. If the confirmation signal is confirmed to be abnormal, a replacement frame is preset, and the replacement frame is obtained through the human-computer interaction interface, and the replacement frame is replaced with the test frame to obtain the safe image; if the confirmation signal is confirmed to be normal, the image to be played is output;

[0077] If the sound abnormality recognition result is abnormal, the sound pre-processing data is passed to the human-computer interaction page, and the confirmation signal is obtained through the human-computer interaction interface, and the confirmation signal is judged. If the confirmation signal is a confirmation of abnormality, a replacement voice is preset, and the replacement voice is obtained through the human-computer interaction interface. The replacement voice is replaced with the sound pre-processing data to obtain the safety image; if the confirmation signal is a confirmation of normal, the image to be played is output.

[0078] It should be noted that, at this time, the image to be played is a safe image.

[0079] Specifically, the decryption server obtains a recognition result according to the encrypted image data through a decryption model, and the player plays the video according to the recognition result, which specifically includes:

[0080] Decrypting the encrypted image data using the SM2 algorithm to obtain hidden watermark information, and comparing the hidden watermark information with the watermark information to obtain the recognition result, wherein the recognition result includes the same and different;

[0081] The player judges the recognition result, and if the recognition result is the same, the encrypted image data is played; if the recognition result is different, the image playback is stopped.

[0082] A public display screen content detection system based on video watermark verification, comprising: a preprocessing module, an identification module, and a decryption module;

[0083] The preprocessing module is used to obtain image data and audio data of the image to be played, obtain image preprocessing data by performing image preprocessing according to the image data, and obtain sound preprocessing data by performing sound preprocessing according to the audio data;

[0084] The recognition module is used to obtain an abnormality recognition result through an abnormality recognition model according to the image preprocessing data, obtain a sound abnormality recognition result through a sound abnormality recognition model according to the sound preprocessing data, and calculate the safety image data through an abnormality processing model according to the abnormality recognition result and the sound abnormality recognition result;

[0085] The decryption module is used to preset watermark information. The encryption server embeds the watermark information into the secure image data through the SM2 algorithm to obtain encrypted image data, and transmits the encrypted image data to the decryption server. The decryption server obtains a recognition result based on the encrypted image data through a decryption model, and the player plays the video based on the recognition result.

[0086] An electronic device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method for detecting content of a public display screen based on video watermark verification as described above is implemented.

[0087] A storage medium containing computer executable instructions, which are used to perform the public display screen content detection method based on video watermark verification as described above when executed by a computer processor.

[0088] The computer storage medium of the embodiment of the present invention may adopt any combination of one or more computer-readable media. The computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples (non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this document, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, device or device.

[0089] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, which carry computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0090] The program code included on the computer readable medium can be transmitted with any appropriate medium, including but not limited to wireless, electric wire, optical cable, RF, etc., or any suitable combination of the above. The computer program code for performing the operation of the present invention can be written in one or more programming languages ​​or their combinations, and the programming language includes object-oriented programming languages-such as Java, Smalltalk, C++, and also includes conventional procedural programming languages-such as "C" language or similar programming languages. The program code can be executed completely on the user's computer, partially on the user's computer, as an independent software package, partially on the user's computer and partially on the remote computer, or completely on the remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (for example, using an Internet service provider to connect through the Internet).

[0091] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment as above, it is not used to limit the present invention. Any technical personnel in this field can make some changes or modify the technical contents disclosed above into equivalent embodiments without departing from the scope of the technical solution of the present invention. However, any simple modification, equivalent change and modification made to the above embodiments according to the technical essence of the present invention without departing from the content of the technical solution of the present invention still fall within the scope of the technical solution of the present invention.

Claims

1. A method for detecting content on a public display screen based on video watermark verification, characterized in that: include: S1: Acquire image data and audio data of a to-be-played image, obtain image preprocessing data by image preprocessing according to the image data, and obtain sound preprocessing data by sound preprocessing according to the audio data; S2: Obtain an abnormality recognition result through an abnormality recognition model according to the image preprocessing data, obtain a sound abnormality recognition result through a sound abnormality recognition model according to the sound preprocessing data, and calculate the safety image data through an abnormality processing model according to the abnormality recognition result and the sound abnormality recognition result; S3: preset watermark information, the encryption server embeds the watermark information into the secure image data through the SM2 algorithm to obtain encrypted image data, the decryption server obtains a recognition result based on the encrypted image data through a decryption model, and the player plays the video based on the recognition result.

2. The method for detecting content on a public display screen based on video watermark verification according to claim 1, characterized in that: The anomaly recognition model includes an image anomaly recognition model and a text anomaly recognition model. The image anomaly recognition model includes: S201: Frame-dividing the image preprocessing data to obtain an image sequence, sequentially traversing the image sequence to obtain a test frame, calculating a feature map using an IMTCNN model based on the test frame, and calculating a feature compensation factor using an operator based on the feature map; S202: Calculate an optimal feature compensation coefficient using a GA algorithm according to the feature compensation factor, and calculate a feature description image according to the optimal feature compensation coefficient and the feature compensation factor; S203: Preset a segmentation template, and segment the feature description image according to the segmentation template to obtain segmentation sub-images; S204: Calculate the pixel value of the segmented sub-image, obtain the coordinates of the segmented sub-image, and splice each pixel value according to the coordinates to obtain a new histogram; S205: According to the new histogram, dimension reduction is performed by using the PCA algorithm to obtain main characteristic factors, and according to the main characteristic factors, an abnormal image recognition result is calculated by using the SVM algorithm.

3. The method for detecting content on a public display screen based on video watermark verification according to claim 2 is characterized in that: The text anomaly recognition model includes: S301: locating text using an SSD algorithm according to the test frame to obtain a text recognition area; S302: performing text recognition according to the text recognition area by using a CRNN algorithm to obtain a text recognition result; S303: performing text preprocessing according to the text recognition result to obtain a preprocessed text, wherein the text preprocessing includes data cleaning, noise removal, and stop word removal; S304: A first-level traversal result is obtained by calculating through a first-level filter according to the preprocessed text, and the first-level traversal result is judged. If the first-level traversal result contains sensitive words, the preprocessed text is passed to the second-level filter, and the second-level filter calculates the text abnormality recognition result according to the preprocessed text; if the first-level traversal result does not contain sensitive words, the calculation is stopped and the text abnormality recognition result is output.

4. The method for detecting content on a public display screen based on video watermark verification according to claim 1, characterized in that: The specific calculation steps of the abnormal sound recognition model are: S401: Obtain an energy spectrum by fast Fourier transform according to the sound preprocessing data, and obtain an amplitude spectrum by removing the phase spectrum according to the energy spectrum; S402: Obtain FBank features through Mel filtering according to the amplitude spectrum, obtain logarithmic features through logarithmic calculation according to the FBank features, and obtain MFCC features through discrete cosine transform according to the logarithmic features; S403: Obtaining speech text data by calculating the MFCC features through a DNN-LSTM model; S404: Calculating a first-level traversal result through a first-level filter according to the voice text data, judging the first-level traversal result, and if the first-level traversal result contains sensitive words, passing the pre-processed text to the second-level filter, and the second-level filter calculates the sound abnormality recognition result according to the voice text data; If the first-level traversal result does not contain sensitive words, the calculation is stopped and the sound abnormality recognition result is output.

5. The method for detecting content on a public display screen based on video watermark verification according to claim 1, characterized in that: The abnormality processing model includes making a judgment according to the abnormality recognition result and the abnormal sound recognition result, and if the abnormality recognition result and the abnormal sound recognition result are normal, outputting the image to be played; If the abnormal identification result is abnormal, the test frame is transmitted to the human-computer interaction page, and a confirmation signal is obtained through the human-computer interaction interface, and the confirmation signal is judged. If the confirmation signal is confirmed to be abnormal, a replacement frame is preset, and the replacement frame is obtained through the human-computer interaction interface, and the replacement frame is replaced with the test frame to obtain the safe image; if the confirmation signal is confirmed to be normal, the image to be played is output; If the sound abnormality recognition result is abnormal, the sound pre-processing data is passed to the human-computer interaction page, and the confirmation signal is obtained through the human-computer interaction interface, and the confirmation signal is judged. If the confirmation signal is a confirmation of abnormality, a replacement voice is preset, and the replacement voice is obtained through the human-computer interaction interface. The replacement voice is replaced with the sound pre-processing data to obtain the safety image; if the confirmation signal is a confirmation of normal, the image to be played is output.

6. The method for detecting content on a public display screen based on video watermark verification according to claim 1, characterized in that: The decryption server obtains a recognition result according to the encrypted image data through a decryption model, and the player plays the video according to the recognition result, specifically including: Decrypting the encrypted image data using the SM2 algorithm to obtain hidden watermark information, and comparing the hidden watermark information with the watermark information to obtain the recognition result, wherein the recognition result includes the same and different; The player judges the recognition result, and if the recognition result is the same, the encrypted image data is played; if the recognition result is different, the image playback is stopped.

7. The method for detecting content on a public display screen based on video watermark verification according to claim 1, characterized in that: The image preprocessing includes filtering noise reduction, histogram equalization, grayscale, data enhancement, light compensation, and geometric correction; the sound preprocessing includes noise reduction, dereverberation, pre-emphasis, and frame segmentation and windowing.

8. A public display screen content detection system based on video watermark verification, characterized in that: include: Preprocessing module, recognition module, decryption module; The preprocessing module is used to obtain image data and audio data of the image to be played, obtain image preprocessing data by performing image preprocessing according to the image data, and obtain sound preprocessing data by performing sound preprocessing according to the audio data; The recognition module is used to obtain an abnormality recognition result through an abnormality recognition model according to the image preprocessing data, obtain a sound abnormality recognition result through a sound abnormality recognition model according to the sound preprocessing data, and calculate the safety image data through an abnormality processing model according to the abnormality recognition result and the sound abnormality recognition result; The decryption module is used to preset watermark information. The encryption server embeds the watermark information into the secure image data through the SM2 algorithm to obtain encrypted image data, and transmits the encrypted image data to the decryption server. The decryption server obtains a recognition result based on the encrypted image data through a decryption model, and the player plays the video based on the recognition result.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the method for detecting content on a public display screen based on video watermark verification as described in any one of claims 1 to 7 is implemented.

10. A storage medium containing computer executable instructions, characterized in that: The computer executable instructions are used to execute the public display screen content detection method based on video watermark verification as described in any one of claims 1-7 when executed by a computer processor.

Images